Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • using estat sbsingle to identify multiple structural breaks

    I'm using Stat 14 with Windows 10 OS.

    I have a time series of quarterly interest rates and I want to test for multiples structural breaks. This is a pick of my data.

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
    clear
    input float qdate double cdi
    147   5.53835376418963
    148  5.110168946328697
    149  4.904248162223812
    150  4.842220311138767
    151 7.7677113279510435
    152  7.191920800029394
    153  5.000530578295859
    154  5.764639419446271
    155  8.086881665616463
    156  8.010016000020027
    157   5.98441457235761
    end
    format %tq qdate
    I used estat sbsingle after a regression of cdi on qdate and found the first break.

    Code:
    regress cdi qdate
    estat sbsingle
    
    ----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
    .................................................. 50
    ..............
    
    Test for a structural break: Unknown break date
    
    Number of obs = 94
    
    Full sample: 1994q4 - 2018q1
    Trimmed sample: 1998q3 - 2014q3
    Estimated break date: 1998q3
    Ho: No structural break
    
    Test Statistic p-value
    -----------------------------------------------
    swald 87.2448 0.0000
    -----------------------------------------------
    Exogenous variables: qdate
    Coefficients included in test: qdate _cons
    My strategy now (and I'm not sure is a good one) is to subsample at the first break and search for next. Them, repeat the procedure and find out all the breaks on my data.

    I just cannot figure it out how to subsample this data because the code below is not working

    Code:
    drop if qdate <= 1998q3
    I have two questions:

    1 - Is my strategy going to work or I'm going to have problems when my time series get too small?
    2 - How can I subsample my data?
Working...
X